Much useful information relevant to elucidation of mechanism of action of nonsteroidal anti-inflammatory drugs (NSAIDs) at the molecular level can be obtained from integrating pharmacokinetic (PK) and pharmacodynamic (PD) data, such data being obtained usually, although not necessarily, in separate studies. Integrating PK and PD data can also provide a basis for selecting clinically relevant dosing schedules for subsequent evaluation in disease models and clinical trials. The principles underlying and uses of PK–PD integration are illustrated in this review for phenylbutazone in the horse and cow, carprofen and meloxicam in the horse, carprofen and meloxicam in the cat and nimesulide in the dog.
In the PK–PD modelling approach for NSAIDs, the PK and PD data are generated (usually though not necessarily) in vivo in the same investigation and then modelled in silico, usually using the integrated effect compartment or indirect response models. Drug effect is classically modelled with the sigmoidal Emax (Hill) equation to derive PD parameters which define efficacy, potency and sensitivity. The PK–PD modelling approach for NSAIDs can be undertaken at the molecular level using surrogates of inhibition of cyclooxygenase (COX) isoforms (or indeed other enzymes e.g. 5-lipoxygenase). Examples are provided of the generation of PD parameters for several NSAIDs (carprofen, ketoprofen, vedaprofen, flunixin and tolfenamic acid) in species of veterinary interest (horse, calf, sheep and goat), which indicate that all drugs investigated except vedaprofen were non-selective for COX-1 and COX-2 in the four species investigated under the experimental conditions used, vedaprofen being a COX-1 selective NSAID. In these studies, plasma concentration was linked to COX inhibitory action in the biophase using an effect compartment model. Data for S-(+)-ketoprofen have been additionally subjected to inter-species modelling and allometric scaling of both PK and PD parameters. For several species values of four PK parameters were highly correlated with body weight, whilst values for PD parameters based on COX inhibition lacked allometric relationship with body weight.
PK–PD modelling of NSAIDs has also been undertaken using clinical end-points and surrogates for clinical end-points in disease models. By measurement of clinically relevant indices in clinically relevant models, data generated for PD parameters have been used to set dosages and dose intervals for evaluation and confirmation in clinical trials. PK–PD modelling of NSAIDs is likely to prove superior to conventional dose titration studies for dosage schedule determination, as it sweeps the whole of the concentration–effect relationship for all animals and therefore permits determination of genuine PD parameters. It also introduces time as a second independent variable thus allowing prediction of dosage interval. Using indirect response models and clinically relevant indices, PD data have been determined for flunixin, phenylbutazone and meloxicam in the horse, nimesulide in the dog and meloxicam in the cat.